Hi, I'm

Jonathan

Computer Science Student & AI/ML Researcher

I study CS at Stanford. I'm interested in all aspects of AI (NLP, computer vision, agent workflows, etc.) and am particularly interested in healthcare applications. More broadly, I also enjoy exploring full-stack development and quantitative finance. I'm also a certified hooper (kind of) and would personally rank myself among the top pickleballers west of the Mississippi. Also find me on the following!

Get In Touch

experience

What I've been up to.

Accenture

LLM Full-stack Engineering Intern

San Francisco, CA

Jun 2025 — Present

Working on agents for enterprise workflow automation at Accenture's Center for Advanced AI.

Stanford Medical / Children's National Hospital

AI/ML Research Assistant

Stanford, CA / Washington, DC

Feb 2024 — Present

Develop and lead research projects ranging from computer vision for ultrasound imaging to LLM applications for patient outcome prediction.

Blueprint Capital / Blueprint Talent Group

Investment Committee Member

Chicago, IL

May 2023 — Present

Lead startup diligence and facilitate investment decisions for VC fund. Developed investment theses, diligence framework, and fund primer.

Stanford Consulting

Vice President

Stanford, CA

Nov 2022 — Present

Arrange team composition and source cases with primarily Fortune 500 companies and PE megafunds. Provide strategic guidance on AI-centered products and investments (gen. AI, computer vision, and LLM-reasoning focused research).

Sparx

Head of Partnerships (Founding Team)

New York, NY

Aug 2023 – May 2024

Led partnerships and business development for a VC-backed fintech/insurtech startup ($3M+ seed round).

Finalyze Capital LLC

Quantitative Trading Intern

Chicago, IL

May 2023 – Sep 2023

Developed and backtested profitable trading algorithms across multiple asset classes.

Awards & Publications

🏆Awards & Honors

National Rank 1 in Public Forum Debate
National Speech and Debate Association (out of 130,000+), 2021, 2022
2x, Second Place at the International Public Policy Forum
(out of 200+ teams from 20+ countries), 2021, 2022
1st Place at Washington Academy of Science Research Convention
2022
Test Scores
  • ACT: 36 (composite)
  • SAT Math II: 800
  • APs (all 5s): Calc BC, Stats, CS, Macro, Micro, English Lang, Comp Gov, NSL Gov, World History

📚Publications & Posters

3D Human Hand Reconstruction Using Gaussian Splatting with Deep Implicit Anatomical Shape PriorsJonathan Wen, Allen Chau, Elijah Song. Stanford University. CS231n Project, 2024.
Determining a Built Infrastructure Vulnerability IndexJennifer Helgeson, Jonathan Wen, Kristin Dan, Elias Nicholson
Segmental Trans-endplate Pedicle Screws do not Induce Spinal Deformity in PigsTaylor R. Johnson, MD MPH, Talissa Oliveira Generoso, MD, Christine Farnsworth, MS, Amishi Jobanputra MS, Jonathan H. Wen, Austin J. Stoner, BS, Vivan Ho, Vidyadhar V. Upasani, MD, John Vorhies, MD
Can AI Replace Patient Reported Outcome Measures? A Pilot StudyJohn S. Vorhies, Stanford University; Arianne T. Salunga, Stanford University; Austin J. Stoner, Stanford Medicine Department of Orthopedic Surgery; Jonathan H. Wen, Stanford University; Hiba Naz, Stanford University; Amishi Jobanputra, Stanford University; Dilan de Silva, Stanford University; Kali R. Tileston, Stanford University

Technical Skills

Programming Languages

PythonC++JavaScriptC#JavaSQLRSTATA

AI/ML Frameworks

PyTorchTensorFlowKerasscikit-learnXGBoostLightGBMFastAIHugging FaceOpenAI APILangChainspaCyNumPypandas

Tools & Platforms

DockerKubernetesAWSGCPAzureGoogle ColabJupyterVS CodeGitMLflowWeights & BiasesStreamlitGradioFlaskFastAPIMongoDBPostgreSQLLinux

projects

A variety of projects I've worked on while trying to learn my passions. You'll find a mix of healthcare AI, computer vision, and web development here, but there isn't a project here that I didn't love devoting many, many hours into.

Laser Lithotripsy Debris Sizer

In Development

Developed a model to conduct size estimations of remaining debris post-operation from laser lithotripsy of kidney stones. Constructed a combination of multiple DCNN models to conduct segmentation and sizing of stone debris, detecting fragments over a specified threshold. Currently being deployed to Stanford Medical Urology Department and undergoing further testing for journal publication.

PyTorchOpenCVComputer VisionDCNNMedical ImagingColab

Agent Patho

Completed

Engineered a multi-step agent to diagnose patients based on symptoms and demographic factors. Optimized the agent's decision-making process by incorporating symptom prioritization algorithms and demographic weighting for improved diagnostic accuracy. Tested the agent with real-world datasets to validate performance and ensure reliable, interpretable outcomes.

PyTorchHugging Face APILangChainMulti-step AgentsHealthcare AI

Email Drafter

Completed

AI-powered email drafting application for personalized outreach to speakers as part of my healthcare leadership TA position. Generates invitation emails using OpenAI API and web search integration. Built with React frontend and FastAPI backend, deployed on Vercel and Render.

ReactFastAPIOpenAI APIPythonVercelRender

Daily Digest

Completed

Automated daily financial digest that provides market insights, economic indicators, and investment opportunities. Updated daily with the latest market analysis and trends.

PythonPandasMachine LearningData AnalysisAPI Integration

3D Hand Reconstruction with Gaussian Splatting

Completed

Developed a novel method for 3D human hand reconstruction from multi-view data by combining 3D Gaussian splatting with a deep implicit anatomical prior learned from the MANO dataset. The approach represents each hand as a set of Gaussian primitives, constrained by a neural network trained on real hand scans to ensure anatomical plausibility. This hybrid framework enables efficient, accurate, and anatomically coherent hand reconstructions, suitable for applications in AR/VR, robotics, and graphics. The method outperforms baseline approaches by integrating model-based priors with data-driven refinement, and is validated on high-resolution multi-view datasets.

PythonOpen3DPyTorch3D Gaussian SplattingMANODeep Learning

Agent Simulation Investing

Completed

Interactive agent-based simulation platform for testing and optimizing investment strategies. Run market simulations and strategy tests directly in your browser!

PythonReinforcement LearningSimulationFinanceAI/ML

Interested in working together or want to see more of my work?

Get In Touch

get in touch

Feel free to reach out!

Location

Stanford, CA

LinkedIn

jon-wen